Research on Arc Fault Classification and Identification of Airborne ACIPDS Based on GA-RF

Abstract

In order to solve the problem of arc fault classification in airborne intelligent power distribution system, an arc fault identification method based on genetic algorithm to optimize random forest was proposed. According to the characteristics of the fault arc, the current characteristics of the arc are first extracted, and then the signal is decomposed into subband signals through the empirical wavelet transform, which overcomes the influence of mode aliasing. Then, the characteristic components reflecting the shape and symmetry of the arc signal composed of energy entropy, sample entropy, root mean square value, waveform factor and skewness factor were extracted and standardized and fused to form the feature vectors of fault classification. Then, the genetic algorithm is used to select the optimal combination of random forest hyperparameters, and the feature vectors are input into the optimized random forest for training to obtain a fault classifier. Finally, the fault classifier was used to classify the test set, with an accuracy of 98.64%, and it was verified in the airborne AC intelligent power distribution system. The results show that the arc fault classification method proposed in this paper can accurately identify the arc fault in the aircraft AC cable network.

Keywords

• AC intelligent power distribution system
• arc faults
• eigenvectors
• random forest