Abstract: In DC electrical systems of more-electric and all-electric aircraft, insulation aging and loose connections can lead to the generation of DC arcs. Compared with AC arcs, DC arcs have no zero-crossing points and are more difficult to extinguish. Sustained arc burning releases significant heat, potentially causing fires and other accidents that endanger safety of onboard equipment and passengers. To propose a DC arc detection method suitable for aircraft electrical systems, an experimental platform is established for DC arc tests in aircraft systems, including resistive and capacitive loads with 28 V and 270 V DC series and parallel arcs. The time-domain and frequency-domain characteristics of DC arcs in aircraft electrical systems are analyzed. Two detection features are proposed: cumulative sum(CUSUM) for change-point detection and sum of autocorrelation coefficients(SACC). CUSUM can detect the instantaneous current fluctuation caused by the generation of the DC arc. Under normal circumstances, the result of SACC is close to 0. When an arc exists, the value of SACC will be relatively large, which can effectively identify the arc. By combining these two features, a CUSUM-SACC-based DC arc detection method is developed. The proposed detection algorithm is tested by experimental data, achieving a detection accuracy of 99.09% at 28 V and 95.73% at 270 V, with a missed detection rate of 0% in both cases.