Search

CN-116540017-B - Series arc fault detection method and system

CN116540017BCN 116540017 BCN116540017 BCN 116540017BCN-116540017-B

Abstract

The invention belongs to the technical field of arc fault detection, and particularly relates to a series arc fault detection method and system, comprising the steps of obtaining a bus current signal; analyzing the obtained current signal, calculating and judging the time domain feature quantity of the current signal when the arc fault occurs to obtain the screened time domain feature quantity, carrying out energy decomposition on the screened time domain feature quantity to obtain the fuzzy entropy of the variation modal component, obtaining the feature evaluation weight by combining the obtained fuzzy entropy, diagnosing the arc fault according to the support vector machine and the obtained feature evaluation weight, and realizing the fault detection of the series arc.

Inventors

  • LIU YUJIAO
  • WANG RUIQI
  • ZHU GUOLIANG
  • MENG FANBO
  • WU XIAOCHUAN
  • ZHANG HUAKUN
  • LIN GUIKE
  • LI JIAPING
  • LI DONGXIN
  • LI GUOLIANG
  • LIN YUQING
  • WANG KUN
  • TANG XIAOGUANG
  • LIN MEIHUA
  • SONG PEIXIN
  • XU XIAOLONG
  • YANG BIN

Assignees

  • 国网山东省电力公司枣庄供电公司
  • 国网山东综合能源服务有限公司

Dates

Publication Date
20260505
Application Date
20230428

Claims (9)

  1. 1. A method of series arc fault detection comprising: Acquiring a bus current signal; Analyzing the obtained current signal, calculating and judging the time domain characteristic quantity of the current signal when arc faults occur, and obtaining the screened time domain characteristic quantity; Performing energy decomposition on the screened time domain feature quantity to obtain fuzzy entropy of variation modal components, and obtaining feature evaluation weights by combining the obtained fuzzy entropy; Diagnosing arc faults according to the support vector machine and the obtained characteristic evaluation weight, and realizing fault detection of the series arc; The decomposition of the time domain feature quantity after screening is carried out, fuzzy entropy of each variable modal component is obtained, the first three variable modal components with the largest fuzzy entropy are selected, current signals are combined and reconstructed to obtain new signal components, the current time domain feature of the new signal components is calculated, and the feature evaluation weight of the current time domain feature of the obtained new signal components is calculated by adopting a Relief-F feature selection algorithm.
  2. 2. A method for detecting a series arc fault as claimed in claim 1 wherein in analyzing the acquired current signal, time domain features of the current signal are calculated and stored, the time domain features including peak-to-peak value, variance, standard deviation, mean, effective value, kurtosis, waveform index, peak factor, pulse factor and skewness.
  3. 3. A method of series arc fault detection as claimed in claim 2 wherein the time domain characteristics of the resulting current signal are monitored, and when the time domain characteristics exceed a set threshold, it is determined that an arc fault is likely to exist, and the time domain characteristics are screened for fault detection.
  4. 4. A method of series arc fault detection as claimed in claim 1 wherein in obtaining the fuzzy entropy of the variant mode components, an improved variant mode decomposition is employed, i.e. a variant mode decomposition is optimised using the australian wild dog algorithm.
  5. 5. The method for detecting arc faults in series as claimed in claim 1, wherein in the process of diagnosing arc faults according to the support vector machine and the obtained characteristic evaluation weight, the integrated time domain characteristic is constructed according to the filtered time domain characteristic and the characteristic evaluation weight, arc fault diagnosis is performed by combining the constructed integrated time domain characteristic and a preset fault diagnosis model, and the fault detection of the series arc is realized by performing secondary fault diagnosis of the obtained fault detection result through the improved support vector machine.
  6. 6. The method of claim 5, wherein the fault diagnosis model uses a random forest algorithm and a modified support vector machine, and the modified support vector machine uses a Australian wild dog algorithm to optimize penalty factors and weight vectors.
  7. 7. A series arc fault detection system, comprising: An acquisition module configured to acquire a bus current signal; The calculation module is configured to analyze the acquired current signals, calculate and judge time domain feature quantities of the current signals when arc faults occur, and obtain screened time domain feature quantities; performing energy decomposition on the screened time domain feature quantity to obtain fuzzy entropy of variation modal components, and obtaining feature evaluation weights by combining the obtained fuzzy entropy; decomposing the filtered time domain feature quantity to obtain fuzzy entropy of each variable modal component, selecting the first three variable modal components with the largest fuzzy entropy, combining and reconstructing the current signal to obtain a new signal component, calculating the current time domain feature of the new signal component, and calculating the feature evaluation weight of the current time domain feature of the obtained new signal component by adopting a Relief-F feature selection algorithm; And the detection module is configured to diagnose arc faults according to the support vector machine and the obtained characteristic evaluation weight and realize fault detection of the series arc.
  8. 8. A computer readable storage medium, on which a computer program is stored, characterized in that the program, when being executed by a processor, implements the steps of the series arc fault detection method as claimed in any one of claims 1-6.
  9. 9. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor implements the steps of the series arc fault detection method according to any one of claims 1-6 when the program is executed.

Description

Series arc fault detection method and system Technical Field The invention belongs to the technical field of arc fault detection, and particularly relates to a series arc fault detection method and system. Background The statements in this section merely provide background information related to the present disclosure and may not necessarily constitute prior art. The low-voltage direct current power supply has the advantages of high efficiency, convenient access to energy storage equipment, high reliability and the like, is widely applied to systems such as photovoltaic power generation, electric automobiles, large data centers and the like, and has wide application prospect in the fields of multi-electric aircrafts, electrified ships, aerospace and direct current distribution and utilization power grids. In the dc power supply system, a dc arc fault occurs when dielectric breakdown, loosening of metal joints, component aging, biting of animals, and the like occur. For alternating current arc, when arc fault occurs, zero-break phenomenon is generated at zero crossing point of current, and direct current arc is different from alternating current arc, the current has no zero crossing point and cannot be naturally quenched, once the arc happens and is not detected in time, the fault can be spread to adjacent circuits, photovoltaic components, power transmission lines, control systems and the like are damaged, and when serious, the continuous burning of the arc can also cause fire accident. According to the inventor, at present, proper characteristic quantity is often selected for fault arc detection and is input into a neural network, but single time domain characteristic is often used as input of a neural network algorithm for the selection of the characteristic quantity, the detection index is single, the precision is low, the instability exists, and the fault information is insufficient. The comprehensive energy system has a plurality of devices, and an electric spark combustion phenomenon often occurs in a circuit, so that once an arc fault occurs, the electric power circuit is greatly damaged, and the safety and the stability of the electric power system are seriously threatened. In the past, for arc fault detection and diagnosis, single characteristic quantity is often selected as an index to judge, so that judgment errors are easy to cause, besides, a power line and the power line are complicated, the position where the arc fault occurs cannot be accurately determined, a great deal of waste of manpower and material resources is caused by one-place investigation, and the efficiency of arc fault detection is extremely low. Disclosure of Invention In order to solve the problems, the invention provides a series arc fault detection method and a series arc fault detection system, which are used for carrying out multidimensional extraction and analysis on arc fault characteristics, avoiding diagnosis errors caused by single index of single extracted characteristics, improving the accuracy of arc fault diagnosis, accurately diagnosing series arc faults in a comprehensive energy system and improving the safety of the comprehensive energy system. According to some embodiments, a first aspect of the present invention provides a method for detecting a series arc fault, which adopts the following technical scheme: A method of series arc fault detection comprising: Acquiring a bus current signal; Analyzing the obtained current signal, calculating and judging the time domain characteristic quantity of the current signal when arc faults occur, and obtaining the screened time domain characteristic quantity; Performing energy decomposition on the screened time domain feature quantity to obtain fuzzy entropy of variation modal components, and obtaining feature evaluation weights by combining the obtained fuzzy entropy; And diagnosing arc faults according to the support vector machine and the obtained characteristic evaluation weight, and realizing fault detection of the series arc. As a further technical definition, in analyzing the acquired current signal, a time domain feature of the current signal is calculated and stored, wherein the time domain feature comprises a peak-to-peak value, a variance, a standard deviation, a mean value, an effective value, a kurtosis, a waveform index, a peak factor, a pulse factor and a skewness. Further, the time domain characteristics of the obtained current signal are monitored, arc faults are judged to be possible when the time domain characteristics exceed a set threshold value, and the time domain characteristics are screened out for fault detection. As a further technical limitation, in the process of obtaining the fuzzy entropy of the variation modal component, an improved variation modal decomposition is adopted, namely, an australian wild dog algorithm is adopted to optimize the variation modal decomposition. As a further technical limitation, decomposing the filtered time domai