針對電力系統中存在的間諧波問題，提出利用子空間分解類演算法中的MUSIC演算法對間諧波檢測。利用信號自相關矩陣的特徵值分解理論，將信號的自相關矩陣分解為信號子空間和雜訊子空間；利用2個子空間的正交性進一步分解雜訊子空間，對其進行變換，構造出基於雜訊子空間分解的特徵多項式；求解該多項式得到信號基波和諧波頻率預估計；最後利用最小二乘法檢測信號的幅值和相位。通過模擬實驗與其他經典演算法比較，結果證明了所提演算法的可行性、高效性和穩定性。

Aimed at the problem of inter-harmonics in power systems, the MUSIC algorithm based on a subspace decomposition algorithm has been proposed. The signal self-correlation matrix is changed into the signal subspace and the noise subspace on the basis of characteristic value decomposition theory of signal autocorrelation matrix. The noise subspace is further decomposed by orthogonality of two subspaces, and gets the characteristic polynomial which is constructed based on the decomposition of the noise subspace and then solves the polynomial to obtain the signal fundamental frequency and harmonic frequency estimation. Finally, amplitude and phase of the signal are detected by the least squares method. Compared with other classical algorithms, the results show that the proposed algorithm is feasible, efficient and stable.