Since the development of smart distribution networks, power electronic interface has been widely used, the harm caused by the inter-harmonics between the more and more attention has been paid. Governance inter- harmonics in the power system, the premise is an accurate analysis of the spectrum between the inter- harmonics. Because Gaussian noise can be completely inhibited by higher-order cumulant, MUSIC algorithm, based on Fourth-order cumulant can be better applied to weak inter-harmonic signal extraction than traditional MUSIC algorithm, to obtain accurate harmonic power spectrum and frequency of the source. Characteristics of the power harmonic are made an accurate analysis by Amplitude calculated using Prony method. This method is verifed by simulation, which can accurate detective the inter-harmonic components and a number of inter- harmonic beside frequency in a short time.