Abstract:In view of the problem that arc fault detection of aviation AC system requires high reliability but generalization ability of the single feature for the arc fault detection method was poor with complex working environment. Arc fault simulation experiments in aviation AC system were carried out. Collecting current data from linear load line when the power frequency was 360Hz, 400Hz and 450Hz . According to the arc characteristics of aviation, a multi-dimension feature detection method of arc fault in aviation AC system was proposed which is fused with waveform distortion , inter harmonic and uncertainty of energy distribution.The support vector machine (SVM) optimized by particle swarm optimization (PSO). Testing set was classified forecast by using classification model. The results showed that, classification accuracy of the designed model was 98.83%。