Distribution line fault waveform feature extraction algorithm based on deep learning

Abstract

The invention discloses a distribution line fault waveform feature extraction algorithm based on deep learning, and relates to the technical field of data processing, and the algorithm comprises the following implementation steps: entering a fault judgment end, collecting the operation parameters of a distribution line in real time, and detecting the island fault of a photovoltaic inverter in real time; performing fault waveform characteristic model comparison on the operating parameters of the power distribution line collected at the current moment in real time, marking unseen fault types, and performing fault judgment; entering a feature monitoring end, detecting whether a fluctuation fault occurs or not in real time by combining environmental parameters, and distinguishing noise interference and thunder and lightning in real time; and entering a multi-dimensional extraction end, and judging whether the fault waveform characteristics are abnormal or not in real time. According to the distribution line fault waveform feature extraction algorithm based on deep learning, fault judgment can be carried out on unseen fault types in real time, faults can be prevented from being mistakenly marked during switching operation and thunder and lightning, and fault waveforms are extracted from multiple angles and multiple positions in combination with waveform features.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and particularly relates to an algorithm for extracting fault waveform features of a distribution line based on deep learning. Background Art

[0002] The extraction of fault waveform features of a distribution line based on deep learning is a technology that uses a neural network model to automatically identify and extract key features of fault signals, such as short circuits, disconnections, arcs, etc., aiming to replace traditional methods that rely on manual experience for threshold judgment or Fourier transform, and improve the accuracy and efficiency of fault detection.

[0003] Currently, with the rapid development of science and technology, there are some deficiencies in the extraction of fault waveform features of a distribution line based on deep learning: 1. The existing technology has weak detection ability for the islanding fault of a photovoltaic inverter, resulting in the inability to perform fault judgment on fault types not seen in the training set in real time during the extraction of fault waveform features, resulting in few real fault samples; 2. For noise interferences such as switch operations and lightning, it is impossible to detect whether there is a fluctuating fault in real time, resulting in easy mislabeling as a fault when switch operations and lightning occur, affecting the accuracy of the extraction of fault fluctuation features of the distribution line; 3. During the operation of the distribution line, it is impossible to combine waveform features to extract fault waveforms in multiple angles and positions in real time, easily resulting in the omission of fault waveform features, further reducing the integrity of the extraction of fault waveform features.

[0004] Therefore, a new algorithm for extracting fault waveform features of a distribution line based on deep learning is proposed to solve the above problems. Summary of the Invention

[0005] The main object of the present invention is to provide an algorithm for extracting fault waveform features of a distribution line based on deep learning to solve the problems proposed in the above background.

[0006] To achieve the above object, the technical solution adopted by the present invention is: an algorithm for extracting fault waveform features of a distribution line based on deep learning, and the method includes the following implementation steps: Step 1: Enter the fault judgment end. By collecting the operating parameters of the distribution line in real time, setting a fault waveform feature model, and detecting the islanding fault of the photovoltaic inverter in real time, compare the operating parameters of the distribution line collected at the current moment with the fault waveform feature model, mark the unseen fault types, and perform fault judgment; Step 2: Enter the feature monitoring end. By combining environmental parameters, collect the noise of the distribution line operation in real time, and detect whether there is a fluctuating fault during switch operations and lightning strikes, and distinguish noise interference and lightning in real time to ensure the accuracy of the extraction of fault fluctuation features of the distribution line; Step 3: Enter the multi-dimensional extraction terminal. By extracting the fault waveform in real time and combining the waveform features in real time to calculate the waveform deviation value from multiple angles and positions, it is judged in real time whether the fault waveform features are abnormal to ensure the integrity of the extraction of fault waveform features.

[0007] The fault judgment terminal includes an operating parameter acquisition unit, an islanding fault detection unit, and a marking judgment unit; The operating parameter acquisition unit is used to collect the operating parameters of the distribution line in real time through a data collector and set a fault waveform feature model according to the standard operating parameters.

[0008] The islanding fault detection unit is used to detect the islanding fault of the photovoltaic inverter in real time through the power line carrier communication device and the frequency relay in combination with the fault waveform feature model, and compare the operating parameters of the distribution line collected at the current moment with the fault waveform feature model in real time. The comparison formula is as follows: ; Where, represents the operating parameter comparison difference corresponding to the th fault waveform feature at the corresponding moment, represents the total number of samples of the operating parameters of the fault waveform feature model, represents the operating parameter of the th fault waveform feature model at time , represents the mean value of the operating parameters of all models at time .

[0009] The marking judgment unit is used to combine and judge whether an islanding fault of the photovoltaic inverter occurs, and mark the fault in real time. When judging, calculate the disturbance frequency during the operation of the distribution line at the current moment. The calculation formula is as follows: ; Where, represents the amplitude of the active power disturbance, represents the disturbance frequency. The disturbance frequency threshold is set to 0.1~5Hz. If the disturbance frequency exceeds 0.1~5 Hz, it is judged that the photovoltaic inverter has an islanding fault. If not, it is judged that the photovoltaic inverter does not have an islanding fault, and the fault data is marked in real time through color identification.

[0010] The feature monitoring terminal includes an environment acquisition unit, a fluctuation fault detection unit, and a fault discrimination and screening unit; The environment acquisition unit is used to collect the operating environment of the distribution line at the current moment through environmental monitoring equipment. The environmental monitoring equipment includes a temperature sensor, a humidity sensor, a leakage current sensor, a partial discharge detector, a voltage sensor, a current sensor, a vibration monitor, and a conductor sag monitor.

[0011] The fluctuation fault detection unit is used to detect the noise during the operation of the distribution line in real time in combination with environmental parameters, and detect whether there is a fluctuation fault in real time for the noise during switch operation and lightning strikes, specifically as follows: Calculate the noise during the operation of the distribution line. The calculation formula is as follows: ; Where, represents the constant related to the noise during switch operation and lightning strikes, represents the vibration frequency in the current environmental conditions, represents the tension during switch operation and lightning strikes in the current environmental conditions, and respectively represent the noise reference values during switch operation and lightning strikes, represents the noise during the operation of the distribution line; Calculate the voltage fluctuation frequency value during switch operation and lightning strikes , and the calculation formula is as follows: ; Where, represents the weight coefficient of the fluctuation frequency band during switch operation and lightning strikes in the operation of the distribution line. Set the voltage fluctuation frequency threshold to 0.75. If the voltage fluctuation frequency value is greater than 0.75, it is determined that there is a fluctuation fault during switch operation and lightning strikes. If not, it is determined that there is no fluctuation fault during switch operation and lightning strikes.

[0012] The fault discrimination and screening unit is used to set the standard fault parameters during switch operation and the standard fault parameters during lightning strikes, and perform fault discrimination and screening in real time in combination with the fault parameters during the operation of the distribution line, specifically as follows: Calculate the actual fault parameters during switch operation. The calculation formula is as follows: ; Where, represents the actual fault parameters under switch operation, represents the voltage drop across the switch, represents the load current during switch operation, represents the resistance during switch operation. If the absolute value of the difference from the standard fault parameter is less than or equal to 0.02, it is determined that the fault is switch operation. If not, it is determined that the fault is not switch operation, and the data determined not to be switch operation is screened for faults through the model; Calculate the actual fault parameters during lightning strikes. The calculation formula is as follows: V GPR =I lightning ⋅Rearth ; Among them, V GPR represents the actual fault parameter during lightning, I lightning represents the lightning current amplitude during lightning, R earth represents the grounding resistance during lightning. Set the safety voltage threshold. If V GPR is greater than the safety voltage threshold, it is determined that the fault is not caused by lightning. Otherwise, it is determined that the fault is caused by lightning. The data indicating that the fault is not caused by lightning is screened for faults through the model.

[0013] The multi-dimensional extraction end includes a progressive mobile fault waveform extraction unit, a multi-dimensional calculation waveform deviation unit, and a real-time judgment of abnormal fault waveform unit; The progressive mobile fault waveform extraction unit is used to capture the fault waveform of the distribution line through a timer and a camera in a mobile manner, and the captured fault waveform of the distribution line is numbered in real time by the timer. The numbering is arranged in ascending order of Arabic numerals.

[0014] The multi-dimensional calculation waveform deviation unit is used to calculate the waveform deviation value of the waveform features in real time in combination with the phases at multiple angles and positions , and the calculation formula is as follows: ; Among them, represents the waveform deviation value of the actual waveform features at different angles and positions, represents the th waveform feature sampling point value of the actual waveform features, represents the th waveform feature sampling point value of the waveform features in the model, represents the dot product sum of the actual waveform features and the waveform features in the model, represents the energy amplitude of the actual waveform, represents the energy amplitude size of the reference waveform.

[0015] The real-time judgment of abnormal fault waveform unit is used to set the standard waveform deviation value of the waveform features at different angles and positions. If the waveform deviation value at different angles and positions is greater than the standard waveform deviation value at different angles and positions, it is determined that the fault waveform is abnormal. If the waveform deviation value at different angles and positions is less than or equal to the standard waveform deviation value at different angles and positions, it is determined that the fault waveform is normal.

[0016] The present invention has the following beneficial effects: 0. In the present invention, by setting a fault judgment end, during the operation of extracting fault waveform features of a distribution line based on deep learning, by comparing the operation parameters of the distribution line collected at the current moment with a fault waveform feature model in real time, it is determined whether a PV inverter islanding fault occurs, and the detection ability for PV inverter islanding faults can be improved. During the extraction of fault waveform features, it is possible to perform fault judgment on unseen fault types in real time, increase real fault samples, and ensure the accuracy of fault waveform detection. 1. In the present invention, by setting a feature monitoring end, during the operation of extracting fault waveform features of a distribution line based on deep learning, by detecting the noise during the operation of the distribution line in real time and distinguishing noise interference and lightning in real time, not only can it be realized to detect whether a fluctuation fault occurs in real time, but also it can be avoided from being mislabeled as a fault during switch operation and lightning occurrence, improving the accuracy of extracting fault fluctuation features of the distribution line. 2. In the present invention, by setting a multi-dimensional extraction end, during the operation of extracting fault waveform features of a distribution line based on deep learning, by calculating the waveform deviation value from multiple angles and positions in real time in combination with waveform features, and judging whether the fault waveform features are abnormal in real time in combination with the waveform deviation value, it is possible to extract fault waveforms from multiple angles and positions in combination with waveform features during the operation of the distribution line, avoiding the omission of fault waveform features and further improving the integrity of extracting fault waveform features. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 is the overall flowchart of an algorithm for extracting fault waveform features of a distribution line based on deep learning according to the present invention; Figure 2 is the schematic structural diagram of the fault judgment end of an algorithm for extracting fault waveform features of a distribution line based on deep learning according to the present invention; Figure 3 is the schematic structural diagram of the feature monitoring end of an algorithm for extracting fault waveform features of a distribution line based on deep learning according to the present invention; Figure 4 is the schematic structural diagram of the multi-dimensional extraction end of an algorithm for extracting fault waveform features of a distribution line based on deep learning according to the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0018] In order to make the technical means, creative features, achieved purposes and effects of the present invention easy to understand, the present invention will be further described below in conjunction with specific embodiments.

[0019] Example 1, please refer to Figures 1 to 2 as shown: An algorithm for extracting fault waveform features of a distribution line based on deep learning includes the following implementation steps: Step 1: Enter the fault judgment terminal. By collecting the operation parameters of the distribution line in real time, set the fault waveform feature model, and detect the islanding fault of the photovoltaic inverter in real time. Compare the operation parameters of the distribution line collected at the current moment with the fault waveform feature model in real time, mark the unseen fault types and enter the fault judgment; Step 2: Enter the feature monitoring terminal. By combining the environmental parameters, collect the noise of the distribution line operation in real time, and detect whether there is a fluctuation fault during switch operation and lightning occurrence. Distinguish the noise interference and lightning in real time to ensure the accuracy of the extraction of the fault fluctuation characteristics of the distribution line; Step 3: Enter the multi-dimensional extraction terminal. By extracting the fault waveform in real time and combining the waveform features from multiple angles and positions to calculate the waveform deviation value in real time, judge whether the fault waveform features are abnormal in real time to ensure the integrity of the extraction of the fault waveform features.

[0020] The fault judgment terminal includes an operation parameter acquisition unit, an islanding fault detection unit, and a marking judgment unit; The operation parameter acquisition unit is used to collect the operation parameters of the distribution line in real time through a data acquisition instrument and set the fault waveform feature model according to the standard operation parameters.

[0021] The islanding fault detection unit is used to detect the islanding fault of the photovoltaic inverter in real time through the power line carrier communication device and the frequency relay in combination with the fault waveform feature model, and compare the operation parameters of the distribution line collected at the current moment with the fault waveform feature model in real time. The comparison formula is as follows: ; Among them, represents the operation parameter comparison difference corresponding to the th fault waveform feature at the corresponding moment, represents the total number of samples of the operation parameters of the fault waveform feature model, represents the th fault waveform feature model's operation parameter at time , represents the mean value of the operation parameters of all models at time .

[0022] The marking judgment unit is used to combine and judge whether there is an islanding fault of the photovoltaic inverter, and mark the fault in real time. When judging, calculate the disturbance frequency during the operation of the distribution line at the current moment. The calculation formula is as follows: ; Among them, represents the active power disturbance amplitude, Denote the disturbance frequency. Set the disturbance frequency threshold to 0.1 - 5 Hz. If the disturbance frequency exceeds 0.1 - 5 Hz, it is determined that the PV inverter has an islanding fault. Otherwise, it is determined that the PV inverter does not have an islanding fault, and the fault data is marked in real time through color identification, and the fault is marked in real time to avoid missing unseen fault types during model comparison, so that the extraction algorithm can improve the detection ability of the PV inverter islanding fault. When extracting the fault waveform characteristics, it can perform fault judgment on unseen fault types in real time, improve the real fault samples, and ensure the accuracy of the fault waveform detection.

[0023] Example 2, please refer to Figure 3 as shown in: Based on Example 1, the feature monitoring end includes an environment acquisition unit, a fluctuation fault detection unit, and a fault discrimination and screening unit; The environment acquisition unit is used to collect the operating environment of the distribution line at the current moment in real time through environmental monitoring equipment. The environmental monitoring equipment includes a temperature sensor, a humidity sensor, a leakage current sensor, a partial discharge detector, a voltage sensor, a current sensor, a vibration monitor, and a conductor sag monitor.

[0024] The fluctuation fault detection unit is used to detect the noise during the operation of the distribution line in real time in combination with environmental parameters, and detect whether there is a fluctuation fault in real time for the noise during switch operation and lightning strikes. Specifically as follows: Calculate the noise during the operation of the distribution line. The calculation formula is as follows: ; Among them, represents the constant related to the noise during switch operation and lightning strikes, represents the vibration frequency in the current environment, represents the tension during switch operation and lightning strikes in the current environment, and respectively represent the noise reference values during switch operation and lightning strikes, represents the noise during the operation of the distribution line; Calculate the voltage fluctuation frequency value during switch operation and lightning strikes , and the calculation formula is as follows: ; Among them, represents the weight coefficient of the fluctuation frequency band during switch operation and lightning strikes in the operation of the distribution line. Set the voltage fluctuation frequency threshold to 0.75. If the voltage fluctuation frequency value is greater than 0.75, it is determined that there is a fluctuation fault during switch operation and lightning strikes. Otherwise, it is determined that there is no fluctuation fault during switch operation and lightning strikes.

[0025] The fault discrimination and screening unit is used to set the standard fault parameters during switch operation and the standard fault parameters during lightning strikes, and combine the fault parameters during the operation of the distribution line to perform real-time fault discrimination and screening. Specifically as follows: Calculate the actual fault parameters during switch operation. The calculation formula is as follows: ; Among them, represents the actual fault parameters under switch operation, represents the voltage drop across the switch, represents the load current during switch operation, represents the resistance during switch operation. If the absolute value of the difference from the standard fault parameters is less than or equal to 0.02, it is determined that the fault is due to switch operation. If not, it is determined that the fault is not due to switch operation, and the data determined not to be due to switch operation is screened for faults through the model; Calculate the actual fault parameters during lightning strikes. The calculation formula is as follows: V GPR =I lightning ⋅R earth ; Among them, V GPR represents the actual fault parameters during lightning strikes, I lightning represents the lightning current amplitude during lightning strikes, and R earth represents the grounding resistance during lightning strikes. Set a safety voltage threshold. If V GPR is greater than the safety voltage threshold, it is determined that the fault is not due to lightning strikes. If not, it is determined that the fault is due to lightning strikes, and the data determined not to be due to lightning strikes is screened for faults through the model. For the noise during switch operation and lightning strikes, it is detected in real time whether there is a fluctuating fault, and the noise interference and lightning are distinguished in real time, so that the noise interference such as switch operation and lightning can be distinguished and screened and judged in real time during the extraction of the fault waveform characteristics of the distribution line, not only realizing the real-time detection of whether there is a fluctuating fault.

[0026] Example 3, please refer to Figure 4 as shown: Based on Example 1, the multi-dimensional extraction end includes a progressive mobile fault waveform extraction unit, a multi-dimensional calculation waveform deviation unit, and a real-time judgment of abnormal fault waveforms unit; The progressive mobile fault waveform extraction unit is used to capture the fault waveforms of the distribution line through a timer and a camera in a mobile manner, and sequence the captured fault waveforms of the distribution line in real time through the timer. The sequence is arranged in ascending order of Arabic numerals.

[0027] The multi-dimensional calculation waveform deviation unit is used to calculate the waveform deviation value of the waveform characteristics in real time by combining the phases at multiple angles and positions , the calculation formula is as follows: ; Among them, represents the waveform deviation value of the actual waveform characteristics at different angles and positions, represents the th waveform characteristic sampling point value of the actual waveform characteristics, represents the th waveform characteristic sampling point value of the waveform characteristics in the model, represents the dot product sum of the actual waveform characteristics and the waveform characteristics in the model, represents the energy amplitude of the actual waveform, represents the energy amplitude of the reference waveform.

[0028] The real-time fault waveform anomaly judging unit is used to set the standard waveform deviation value of the waveform characteristics at different angles and positions. If the waveform deviation value at different angles and positions is greater than the standard waveform deviation value at different angles and positions, it is judged that the fault waveform is abnormal. If the waveform deviation value at different angles and positions is less than or equal to the standard waveform deviation value at different angles and positions, it is judged that the fault waveform is normal. Combining the waveform deviation value, it can judge whether the fault waveform characteristics are abnormal in real time, so that the extraction algorithm can further ensure the integrity of the extraction of the fault waveform characteristics. When the distribution line is operating, it can extract the fault waveform from multiple angles and positions in combination with the waveform characteristics.

[0029] In the present invention, for a fault waveform feature extraction algorithm for a distribution line based on deep learning, first, configure a control terminal server for fault waveform feature extraction of the distribution line based on deep learning. Enter the fault judgment end. Real-time collect the operating parameters of the distribution line through a data collector, and set a fault waveform feature model according to the standard operating parameters. Combine the fault waveform feature model to detect the islanding fault of the photovoltaic inverter in real time. Compare the operating parameters of the distribution line collected at the current moment with the fault waveform feature model in real time to determine whether an islanding fault of the photovoltaic inverter occurs, and mark the fault in real time to avoid missing unseen fault types during model comparison, so that the extraction algorithm can improve the detection ability for the islanding fault of the photovoltaic inverter. When extracting fault waveform features, it can judge unseen fault types in real time, improve real fault samples, and ensure the accuracy of fault waveform detection; Enter the feature monitoring end. Real-time collect the environmental parameters during the operation of the distribution line, and combine the environmental parameters to detect the noise during the operation of the distribution line in real time. Detect whether a fluctuation fault occurs in real time for the noise during switch operation and lightning, and distinguish noise interference and lightning in real time, so that noise interference such as switch operation and lightning can be distinguished and screened and judged in real time during the extraction of the fault waveform features of the distribution line, not only realizing the real-time detection of whether a fluctuation fault occurs, but also avoiding being mislabeled as a fault during switch operation and lightning, and improving the accuracy of the extraction of the fault fluctuation features of the distribution line; By adopting a progressive mobile method to extract the fault waveform in real time, and combining the waveform deviation value calculated from multiple angles and positions of the waveform features in real time, and combining the waveform deviation value to judge whether the fault waveform features are abnormal in real time, so that the extraction algorithm can further ensure the integrity of the extraction of the fault waveform features. When the distribution line is operating, it can extract the fault waveform from multiple angles and positions in combination with the waveform features, avoid missing the fault waveform features, and further improve the integrity of the extraction of the fault waveform features.

[0030] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art of this industry should understand that the present invention is not limited by the above embodiments. What is described in the above embodiments and the specification only illustrates the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention will have various changes and improvements, and these changes and improvements all fall within the scope of the present invention claimed. The scope of protection claimed by the present invention is defined by the appended claims and their equivalents.

Claims

1. A fault waveform feature extraction algorithm for distribution lines based on deep learning, characterized in that, It includes the following implementation steps: Step 1: Enter the fault judgment terminal. By collecting the operation parameters of the distribution line in real time, set the fault waveform feature model, and detect the islanding fault of the photovoltaic inverter in real time. Compare the operation parameters of the distribution line collected at the current moment with the fault waveform feature model in real time, mark the unseen fault types, and conduct fault judgment; Step 2: Enter the feature monitoring terminal. By combining the environmental parameters, collect the noise of the distribution line operation in real time, and detect whether there is a fluctuation fault during switch operation and lightning occurrence. Distinguish the noise interference and lightning in real time to ensure the accuracy of the extraction of the fault fluctuation characteristics of the distribution line; Step 3: Enter the multi-dimensional extraction terminal. By extracting the fault waveform in real time, and combining the waveform features to calculate the waveform deviation value from multiple angles and positions in real time, judge whether the fault waveform features are abnormal in real time to ensure the integrity of the extraction of the fault waveform features.

2. The algorithm according to claim 1, characterized in that: The fault judgment terminal includes an operation parameter acquisition unit, an islanding fault detection unit, and a marking judgment unit; The operation parameter acquisition unit is used to collect the operation parameters of the distribution line in real time through a data acquisition instrument, and set the fault waveform feature model according to the standard operation parameters.

3. The algorithm according to claim 2, wherein: The islanding fault detection unit is used to detect the islanding fault of the photovoltaic inverter in real time through power line carrier communication equipment and frequency relays in combination with the fault waveform feature model, and compare the operation parameters of the distribution line collected at the current moment with the fault waveform feature model in real time. The comparison formula is as follows: ； Among them, represents the comparison difference of the operating parameters corresponding to the th fault waveform feature at the corresponding moment, represents the total number of samples of the operating parameters of the fault waveform feature model, represents the th operating parameter of the fault waveform feature model at time , represents the mean value of the operating parameters of all models at time .

4. The algorithm according to claim 3, characterized in that: The marking judgment unit is used to judge whether there is an islanding fault of the photovoltaic inverter in combination, and mark the fault in real time. When judging, calculate the disturbance frequency during the operation of the distribution line at the current moment. The calculation formula is as follows: ； Among them, represents the amplitude of the active power disturbance, represents the disturbance frequency. The disturbance frequency threshold is set to 0.1 - 5 Hz. If the disturbance frequency exceeds 0.1 - 5 Hz, it is determined that the PV inverter has an islanding fault. Otherwise, it is determined that the PV inverter has no islanding fault, and the fault data is marked in real time through color identification.

5. The algorithm according to claim 1, wherein: The feature monitoring terminal includes an environment acquisition unit, a fluctuation fault detection unit, and a fault discrimination and screening unit; The environment acquisition unit is used to collect the operation environment of the distribution line at the current moment through environmental monitoring equipment. The environmental monitoring equipment includes a temperature sensor, a humidity sensor, a leakage current sensor, a partial discharge detector, a voltage sensor, a current sensor, a vibration monitor, and a conductor sag monitor.

6. The algorithm according to claim 5, wherein: The fluctuation fault detection unit is used to detect the noise during the operation of the distribution line in real time by combining the environmental parameters, and detect whether there is a fluctuation fault in real time for the noise during switch operation and lightning occurrence. Specifically as follows: Calculate the noise during the operation of the distribution line. The calculation formula is as follows: ； Among them, represents a constant related to noise during switch operation and lightning strikes, represents the vibration frequency in the current environmental conditions, represents the tension during switch operation and lightning strikes in the current environmental conditions, and respectively represent the noise reference values during switch operation and lightning strikes, represents the noise during the operation of the distribution line; Calculate the voltage fluctuation frequency values during switch operations and lightning strikes , and the calculation formula is as follows: ； Among them, represents the weight coefficient of the fluctuation frequency band during switch operation and lightning strike in the operation of the distribution line. The voltage fluctuation frequency threshold is set to 0.

75. If the voltage fluctuation frequency value is greater than 0.75, it is determined that there is a fluctuation fault during switch operation and lightning strike. Otherwise, it is determined that there is no fluctuation fault during switch operation and lightning strike.

7. The algorithm according to claim 6, wherein: The fault discrimination and screening unit is used to set the standard fault parameters during switch operation and the standard fault parameters during lightning occurrence, and conduct fault discrimination and screening in real time by combining the fault parameters during the operation of the distribution line. Specifically as follows: Calculate the actual fault parameters during switch operation. The calculation formula is as follows: ； Among them, represents the actual fault parameter under switch operation, represents the voltage drop across the switch, represents the load current during switch operation, represents the resistance during switch operation. If the absolute value of the difference from the standard fault parameter is less than or equal to 0.02, it is determined that the fault is due to switch operation; otherwise, it is determined that the fault is not due to switch operation, and the data indicating that the fault is not due to switch operation is screened for faults through the model. Calculate the actual fault parameters during lightning occurrence. The calculation formula is as follows: V GPR = I lightning ⋅ R earth ; Among them, V GPR represents the actual fault parameter during lightning strike, I lightning represents the lightning current amplitude during lightning strike, R earth represents the grounding resistance during lightning strike. Set the safety voltage threshold. If V GPR is greater than the safety voltage threshold, it is determined that the fault is not caused by lightning strike. Otherwise, it is determined that the fault is caused by lightning strike. The data indicating that the fault is not caused by lightning strike is screened for faults through the model.

8. The algorithm according to claim 1, characterized in that: The multi-dimensional extraction terminal includes a progressive mobile fault waveform extraction unit, a multi-dimensional waveform deviation calculation unit, and a real-time abnormal fault waveform judgment unit; The progressive mobile fault waveform extraction unit is used to capture the fault waveform of the distribution line in a mobile manner through a timer and a camera, and arrange the captured fault waveforms of the distribution line in sequence in real time through the timer. The sequence arrangement is carried out in ascending order of Arabic numerals.

9. The algorithm according to claim 8, wherein: The multi-dimensional calculation waveform deviation unit is used to combine the waveform deviation values of the phase calculation waveform features at multiple angles and multiple positions in real time , and the calculation formula is as follows: ； Among them, represents the waveform deviation value of the actual waveform characteristics at different angles and positions, represents the th waveform characteristic sampling point value of the actual waveform characteristics, represents the th waveform characteristic sampling point value of the waveform characteristics within the model, represents the dot product sum of the actual waveform characteristics and the waveform characteristics within the model, represents the energy amplitude of the actual waveform, represents the magnitude of the energy amplitude of the reference waveform.

10. The algorithm according to claim 9, wherein: The real-time fault waveform abnormality judgment unit is used to set the standard waveform deviation value of waveform features at different angles and different positions. If the waveform deviation value at different angles and different positions is greater than the standard waveform deviation value at different angles and different positions, it is judged that the fault waveform is abnormal. If the waveform deviation value at different angles and different positions is less than or equal to the standard waveform deviation value at different angles and different positions, it is judged that the fault waveform is normal.

Images