Abstract: In order to solve the harmonic recovery problem in the background of α noise，a normalized cyclic
correlation combined with multiple signal classification algorithm is proposed． It consists of two proposed
multiple signal classification algorithms ( MUSIC: Multiple Signal Classification ) : sample space MUSIC
algorithm and feature space MUSIC algorithm． The two MUSIC algorithms make full use of the signal subspace
and the noise subspace to perform spectral peak search in the spatial domain to find the harmonic frequeacy．
This algorithm can estimate the harmonic signal frequency and improve the accuracy of harmonic estimation．
The computer simulation results show that the two algorithms can effectively estimate the frequency of the
harmonic signal，and the effect is better than the original fractional lower moment and its derived fractional loworder
statistic，and effectively solve the non-integer operator． The resulting phase distortion problem has broad
application prospects．

Key words:  ,  , α noise, multiple signal classification( MUSIC) algorithm, harmonic recovery