Arc fault detection using fuzzy logic

• US 8,054,592 B2
• Filed: 10/03/2008
• Issued: 11/08/2011
• Est. Priority Date: 10/03/2008
• Status: Active Grant

• Current Assignee: ABB Schweiz AG (ABB Limited)
• Original Assignee: General Electric Company
• Inventors: Rivers, Cecil Jr.
• Primary Examiner(s): Fureman; Jared
• Assistant Examiner(s): Bauer; Scott
• Application Number:
  US12/245,415
• Publication Number:
  US 20100085669A1
• Time in Patent Office: 1,131 Days
• Field of Search: 361/42
• US Class Current: 361/42
• CPC Class Codes: G01R 31/1272 of cable, line or wire insulation, e.g. using partial discharge measurements locating faults in cables G01R31/083
• CPC Class Codes: H02H 1/0015 Using arc detectors
• CPC Class Codes: H02H 1/0092 concerning the data processing means, e.g. expert systems, neural networks

Abstract

An arc fault detector which includes a filter which receives an input signal and filters the input signal, an amplifier which amplifies a signal output from the filter, an analog-to-digital converter disposed to receive the amplified signal from the amplifier and convert the amplified signal into a digital signal, and a processing unit responsive to computer executable instructions when executed thereon that receives samples of data associated with the digital signal and performs an arc detection algorithm on the data using fuzzy logic.

Claims

1. An arc fault detector comprising:

a filter which receives an input signal and filters the input signal;

an amplifier which amplifies a signal output from the filter;

an analog-to-digital converter disposed to receive the amplified signal from the amplifier and convert the amplified signal into a digital signal; and

a processing unit responsive to computer executable instructions when executed thereon that receives samples of data associated with the digital signal and performs an arc detection algorithm on the data using fuzzy logic and to determine a spread and an amount of dead time of the data and detect arcing fuzzy logic based on the spread and the amount of dead time determined.

2. The arc fault detector of claim 1, further comprising a current sensor which measures current on a conductor of an electrical circuit, the current sensor providing a signal representative of the input signal to the filter.

3. The arc fault detector of claim 1, wherein current on a conductor of an electrical circuit is measured via a bimetal, the bimetal providing a signal representative of the input signal to the filter.

4. The arc fault detector of claim 2, wherein the filter comprises an electrical circuit that removes a fundamental line frequency from the input signal.

5. The arc fault detector of claim 1, wherein the processing unit is further responsive to the computer executable instructions to determine the spread and the amount of dead time of the data per one cycle.

6. The arc fault detector of claim 1, wherein the processing unit is further responsive to the computer executable instructions to determine the spread and the amount of dead time of the data per specified intervals of a cycle.

7. The arc fault detector of claim 5, wherein the processing unit is further responsive to the computer executable instructions to repeatedly perform the arc detection algorithm over a specified period of time, and generate a trip signal when it is determined that a predetermined number of arcs have occurred in the specified period of time.

8. An arc fault detection method comprising:

capturing samples of electronic data representative of measured current on a conductor of an electrical circuit; and

performing an arc fault detection algorithm on the captured data and using fuzzy logic to detect arc faults, wherein performing an arc fault detection algorithm comprises;

determining an absolute value of the captured data;

converting the absolute value of the captured data into a bit level based on a predetermined threshold;

determining a rolling average of the captured data based on the bit level;

calculating a spread and an amount of dead time of the captured data; and

detecting arcing using fuzzy logic based on the spread and the amount of dead time calculated.

9. The arc fault detection method of claim 8, wherein capturing samples of electronic data comprises:

measuring current on the conductor of the electrical circuit;

filtering a fundamental line frequency from the measured current;

amplifying a signal associated with the measured current; and

offsetting a DC component of the amplified signal.

10. The arc fault detection method of claim 9, wherein converting the absolute value of the captured data into a bit level based on a predetermined threshold comprises:

converting the captured data into a bit level of 1 when an amplitude of the captured data is over the predetermined threshold; and

converting the captured data into a bit level of 0 when the amplitude of the captured data is below the predetermined threshold.

11. The arc fault detection method of claim 10, wherein calculating the spread comprises determining an average of a number of times the bit level is equal to 1, and calculating the amount of dead time comprises determining a number of times the bit level is equal to 0.

12. The arc fault detection method of claim 11, wherein detecting arcing using fuzzy logic comprises:

calculating membership degrees based on predetermined membership functions for the spread and the amount of dead time calculated; and

calculating an arcing membership degree and determining an arcing level based on the calculated membership degrees for the spread and the amount of dead time calculated and a predetermined arcing membership function;

generating a shape on the predetermined arcing membership function using the calculated arcing membership degree;

calculating a center of mass of the shape on the predetermined arcing membership function; and

comparing the calculated center of mass to specified thresholds to determine whether arcing is present.

13. The arc fault detection method of claim 8, further comprising repeatedly capturing samples of data and performing the arc fault detection algorithm using fuzzy logic on the captured data over a specified period of time, and generating a trip signal to trip a circuit breaker when it is determined that a predetermined number of arcs have occurred in the specified time period.

14. A non-transitory storage medium encoded with machine-readable computer program code executable by computer for arc fault detection, the non-transitory storage medium including instructions for execution by the computer to implement an arc fault detection method, the method comprising:

capturing samples of electronic data representative of measured current on a conductor of an electrical circuit; and

performing at the processing circuit an arc fault detection algorithm on the captured data using fuzzy logic to detect and protect against arc faults, wherein performing an arc fault detection algorithm comprises;

determining an absolute value of the captured data;

converting the absolute value of the captured data into a bit level based on a predetermined threshold;

determining a rolling average of the captured data based on the bit level;

calculating a spread and an amount of dead time of the captured data; and

detecting arcing using fuzzy logic based on the spread and the amount of dead time calculated.

15. The storage medium of claim 14, wherein converting the absolute value of the captured data into a bit level based on a predetermined threshold comprises:

converting the captured data into a bit level of 1 when an amplitude of the captured data is over the predetermined threshold; and

converting the captured data into a bit level of 0 when the amplitude of the captured data is below the predetermined threshold.

16. The storage medium of claim 15, wherein calculating the spread comprises determining an average of a number of times the bit level is equal to 1, and calculating the amount of dead time comprises determining a number of times the bit level is equal to 0.

17. The storage medium of claim 16, wherein detecting arcing using fuzzy logic comprises:

calculating membership degrees based on predetermined membership functions for the spread and the amount of dead time calculated; and

calculating an arcing membership degree and determining an arcing level based on the calculated membership degrees for the spread and the amount of dead time calculated and a predetermined arcing membership function;

generating a shape on the predetermined arcing membership function using the calculated arcing membership degree;

calculating a center of mass of the shape on the predetermined arcing membership function; and

comparing the calculated center of mass to specified thresholds to determine whether arcing is present.

18. The storage medium of claim 14, wherein the arc fault detection method further comprises repeatedly capturing samples of data and performing the arc fault detection algorithm using fuzzy logic on the captured data over a specified period of time, and generating a trip signal to trip a circuit breaker when it is determined that a predetermined number of arcs have occurred in the specified time period.