Arc fault detection in aircraft AC power systems using coefficient of variation

Arc fault is one of the significant factors affecting the flight safety of aircraft. Arc faults may directly cause damage to electrical equipment, interrupt power supply, or ignite fires, posing serious risks to flight safety, it is particularly important to strengthen the detection of arc faults. When a single feature is used to identify series arc faults, the fixed threshold obtained solely through subjective experience cannot adapt to the fault detection requirements of different loads. The fault detection performance of some loads under fixed thresholds is subpar. To address this issue, this paper proposes an adaptive threshold fault detection method based on the coefficient of variation features (CV) which combines standard deviation and mean of the frequency of current. Firstly, the CV of multiple load current frequencies are extracted. Then the optimal distribution of CV is determined as double Weibull distribution by the minimum error sum of squares criterion. The adaptive threshold is established based on the double Weibull distribution and its confidence intervals. The results show that the proposed method achieves an average fault identification accuracy of 98.8 %, which is improved by 4.55 %, 15.3 % and 17.05 % compared to the P-value of Levene test, skewness and correlation coefficient respectively.