| Title |
A Study on an Arc Detection Device by Noise Learning Algorithm |
| Authors |
곽동걸(Dong-Kurl Kwak) ; 정민상(Min-Sang Jung) ; 백원종(Won-Jong Baek) ; 류진규(Jin-Kyu Ryu) |
| DOI |
https://doi.org/10.5370/KIEE.2023.72.10.1306 |
| Keywords |
Electric fire; Arc; Noise learning algorithm; AFCI; Noise counter comparator |
| Abstract |
It is very high that the proportion of fires caused by electrical factors in Korea. In particular, fires caused by arc faults account for most of the total electric fire. However, the prevention of arc faults is insufficient in Korea. Recently, arc detection devices have been developed to analyze the amplitude and frequency of arc_type noise signal. However, since these devices must count all noise signals or analyze frequency components in real time, harmless noise components similar to arc_type noise signals are also recognized as arc signals, resulting in malfunction and reliability degradation in arc fault detection. This paper proposes a new arc detection device to improve for the problems of existing arc detectors. The proposed arc detection device has an algorithm that periodically learns general non-risk noise generated in the customer's power environment, and then compares and analyzes it with the learned noise in the event of an arc fault. As a result, the proposed arc detection device ignores harmless noise components and detects only high-risk arc_type noise components, further enhancing the reliability and precision of arc detection.
In addition, the proposed arc detection device verifies the validity of theoretical analysis and the practicality of the device through several actual measurements in various power environments. |