“…However, the harmonic components (1 ± 2 ) are very close to the power frequency components , so it adds many difficulties to detect the components of fault feature frequency, which are prone to be submerged in the strong supply power frequency component in the FFT spectrum. To solve this problem, many researchers focused on studying the zoom-FFT, marginal spectrum [1], high-resolution spectrum [11], higher order spectral analysis [12], wavelet analysis [13,14], Kalman filters, the self-adaptive filter [1], and so forth.…”