This paper presents the application of fuzzy ARTMAP to investigate a long outstanding problem of arc type fault detection in electric systems, distribution systems, process industries and mining. With the recent increase in the use of nonlinear type loads, power semiconductor converters and possible harmonic series and parallel resonance conditions in power system and the consequent increase in the total harmonic distortion (THD), it is very difficult to use a simple relay based only a single harmonic component for HIF arc fault detection. In this paper, a collection of dominant key harmonic signals forms the basis for a dasiafeaturepsila or dasiadiagnosticpsila vector instead of a single dominant harmonic signal. Fast Fourier transform (FFT) is used to transform the HIF arc faults from the time domain to the frequency domain. A novel Feature Vector comprising the 3rd and 5th harmonics of the voltage, current and power signals is developed.