Abstract:

Aiming at the difficulties in measuring aviation series arc fault in AC system problems caused by wider spectrum range, a kind of analysis method of arc fault spectral characteristic was put forward by the combination of wavelet packet, information entropy and fast Fourier transform. Firstly, a series of arc fault simulation experiments of different loads in aviation AC system were carried out. Secondly, the best mother wavelet and decomposition level were chosen and characteristic band of the series arc fault current signal before and after arc burning was extracted by using wavelet packet technology, singularity entropy theory and negentropy theory. Finally, interharmonic feature of the arc fault current signal before and after arc burning was compared on characteristic band. The results showed that interharmonic feature in the characteridtic frequence band is more advantageous for the detection of arc fault than that over the entire frequency range.

Key words: Arc fault, frequency band, wavelet packet, information entropy, inter harmonic