Electrical series arc detection using continuous wavelet transforms

Electrical and Electronic Engineering not elsewhere classified

Khan, Maryam ; West, Sam

CSIRO

DOI  :  10.25919/5c9a68b111c6c
RP-ID  :  EP19854

学科分类：地球科学(综合)

澳大利亚|英语

来源: CSIRO Research Publications Repository

【 摘 要 】

Electrical arcing is a leading cause of fires in residential, commercial and industrial buildings. Electrical arcing was the cause of 16% of all residential building fires in the US between 2014-2016, resulting in 2,695 deaths, 12,000 injuries and $7 billion in property loss [1]. Similarly, in large non-residential building fires, arcing was the cause of 15% of fires, 90 deaths and $2.4 billion in property loss per year [2]. In the UK, electrical fires, including those caused by arcing, were responsible for 37% of all residential fire-related fatalities between 2017 and 2018, causing 283 deaths [3].Other electrical problems, such as such as over-current and earth leakage faults are dealt with by common electrical protection devices like the stand-alone Arc Fault Circuit Interrupter (AFCIs). While such devices can be purchased commercially, the uptake has been slow because, installation is not mandatory in most jurisdictions and industries and because of the high rates of false-alarm trips when used with certain appliances.This paper describes the detection of electric arcs using advanced spectral decomposition methods on high frequency samples of current and voltage waveforms. The method can be made very sensitive to individual arc events and tuned to suit specific applications and load types via adjustable thresholds.