Arc　fault　is　a　prevalent　phenomenon　in　low-voltage　residential　power　systems　and　the　main　cause　of　electrical　fires.　This　paper　explores　the　fire　risks　associated　with　arc　faults　and　employs　standards　from　arc　fault　detection　devices　(AFDD)　to　create　an　arc　fault　experimental　system　complemented　by　an　integrated　multi-sensor　system　for　ignition　related　data　collection.　The　objective　is　to　synthetically　examine　the　dynamic　characteristics　and　ignition　mechanisms　of　arc　faults.　This　paper　investigates　the　generation　mechanisms,　waveforms,　and　energy　characteristics　of　arc　faults,　including　waveform　distinctions　across　different　load　types　and　patterns　of　energy　variation　during　arcing.　Data　from　the　multi-sensor　system　become　the　cornerstone　of　the　correlation　analysis　between　the　current,　energy,　and　ignition　phenomena,　facilitating　the　construction　of　an　ignition　probability　model　based　on　maximum　likelihood　estimation.　The　reliability　of　the　model　is　verified　across　various　datasets　with　a　95　%　confidence　interval,　offering　a　quantitative　fire-inducing　risk　assessment　of　arc　faults.　Furthermore,　the　paper　assesses　the　AFDD　standards＇　break　time　limits　and　evaluates　the　performance　of　seven　AFDD　prototypes　based　on　the　proposed　model,　demonstrating　the　effectiveness　of　this　technology　in　mitigating　the　risks　of　fires　induced　by　arc　faults.　This　study　systematically　analyzes　the　behaviors　and　ignition　mechanisms　of　arc　faults,　thereby　providing　scientific　evidence　and　data　that　support　the　enhancement　of　arc　fault　detection　technologies　and　the　development　of　fire　prevention　strategies.