Sub-harmonic arc fault detection system and method
Embodiments of the invention provide systems and methods for detecting a sustained arc in an electrical system. Over a time period, current and voltage data for a load signal are collected, from which spectral information is extracted. The spectral information has a frequency component and an amplitude component. The load signal is processed to remove a line frequency signature from the spectral information. One or more sub-harmonic frequency bands are extracted from the load signal and analyzed to determine the presence of an arc signature therein. The sub-harmonic frequency bands may be centered on frequencies corresponding to an integer number of half line cycle periods. Analysis of the sub-harmonic frequency bands can include detecting the peak amplitudes of the rectified sub-harmonic frequency bands to obtain an indicator signal and determining if the indicator signal exceeds a threshold that indicates the presence of the sustained arc.
1. A method of detecting a sustained arc in an electrical system, the method comprising:
collecting over a predetermined time period a load signal by sampling in the time domain and subsequently extracting spectral information comprising a frequency component and an amplitude component;
determining a line current frequency of the electrical system, the line current frequency being a real-time frequency of a line current;
removing a line frequency signature from the spectral information, the line frequency signature comprising the line current frequency of the electrical system and all frequencies above the line current frequency;
subsequent to removing the line frequency signature, extracting at least one sub-harmonic frequency band from the spectral information, wherein each of the at least one sub-harmonic frequency band contains only frequencies that are less than the determined line current frequency; and
analyzing the at least one sub-harmonic frequency band to determine whether an arc signal is present.
2. The method of claim 1 wherein removing the line frequency signature comprises:
tracking an input signal that contains the line frequency signature;
extracting the line current frequency from the input signal as the input signal is tracked; and
using the line current frequency to remove the line frequency signature from the spectral information.
3. The method of claim 2 wherein extracting the line current frequency from the input signal comprises applying an adaptive notch filter to the input signal.
4. The method of claim 1 wherein extracting the at least one sub-harmonic frequency band comprises applying at least one bandpass filter to the spectral information, the at least one bandpass filter comprising a pass-band around a frequency corresponding to twice the line current frequency divided by an integer greater than or equal to three.
5. The method of claim 1 wherein extracting the at least one sub-harmonic frequency band comprises dividing the at least one sub-harmonic frequency band into at least one time segment.
6. The method of claim 5 wherein the at least one time segment is between 0.5 and 3 seconds.
7. The method of claim 1 wherein analyzing the at least one sub-harmonic frequency band to determine whether an arc signal is present comprises:
detecting an amplitude peak in at least one sub-harmonic frequency band to obtain an indicator signal; and
determining if the indicator signal exceeds a threshold that indicates whether the sustained arc is present.
8. The method of claim 7 wherein the amplitude peak exceeds a baseline amplitude by an amount and for a duration adapted from a known arc signature, the pattern being determined by laboratory testing of a type of electrical system.
9. The method of claim 7 wherein analyzing the at least one sub-harmonic frequency band to determine whether an arc signal is present further comprises deriving at least one historical value for the spectral information, and wherein determining if the indicator signal exceeds a threshold comprises determining how long the indicator signal exceeds the at least one historical value.
10. The method of claim 7 wherein analyzing the at least one sub-harmonic frequency band to determine whether an arc signal is present further comprises deriving at least one historical value for the spectral information, and wherein determining if the indicator signal exceeds a threshold comprises determining by how much the indicator signal exceeds the at least one historical value.
11. A method of detecting a sustained are in an electrical system, the method comprising:
collecting over a predetermined time period a load signal;
extracting spectral information from the load signal, the spectral information comprising a frequency component and an amplitude component;
removing a line frequency signature from the spectral information, the line frequency signature comprising a measured line frequency;
removing from the spectral information all frequencies above the measured line frequency;
applying to the spectral information a plurality of bandpass filters to obtain a plurality of sub-harmonic frequency bands, wherein the plurality of sub-harmonic frequency bands each have only lower frequencies than the measured line frequency;
dividing the plurality of sub-harmonic frequency bands into a plurality of time segments;
detecting an amplitude peak within each sub-harmonic frequency band in the plurality of time segments; and
determining from the amplitude peaks if the arc signature is present in the plurality of time segments.
12. The method of claim 11 wherein the plurality of time segments are each between about 0.5 and 3 seconds long.
13. The method of claim 11 wherein determining the amplitude peak in each sub-harmonic frequency band produces an indicator signal, and wherein determining if the arc signature is present in the plurality of time segments comprises determining if the indicator exceeds at least one historical value of the amplitude component of the spectral information.
14. The method of claim 13 wherein determining if the arc signature is present in the plurality of time segments further comprises determining for how long the indicator exceeds the at least one historical value.
15. The method of claim 11 further comprising:
tracking an input signal that contains the line frequency signature; and
extracting the measured line frequency from the input signal.
16. The method of claim 15 wherein extracting the measured line frequency from the input signal comprises applying an adaptive notch filter to the input signal.
17. A method of detecting a sustained arc in an electrical system, the method comprising:
collecting over a predetermined time period a load signal;
extracting spectral information from the load signal, the spectral information comprising a frequency component and an amplitude component;
extracting a measured line frequency from an input signal;
removing the measured line frequency from the spectral information;
applying to the spectral information a plurality of bandpass filters to obtain a plurality of sub-harmonic frequency bands, wherein the sub-harmonic frequency bands each have only lower frequencies than the measured line frequency;
dividing the plurality of sub-harmonic frequency bands into a plurality of time segments; and
determining if an amplitude peak within each time segment exceeds a threshold that indicates whether the sustained arc is present.
18. The method of claim 17 further comprising deriving at least one historical value for the spectral information, and wherein determining if the indicator signal exceeds a threshold comprises determining how long the indicator signal exceeds the at least one historical value.
19. The method of claim 17 wherein each of the bandpass filters comprises a pass-band around a frequency corresponding to twice the line current frequency divided by an integer greater than or equal to three.