Abstract: In experimental modal analysis (EMA), the average approach is generally used to process the random noise in vibration signals. In many cases, the problem of harmonic components, which may be caused by periodic excitation to structures, is difficult to solve by the average approach, and this can lead to erroneous modal identification. In this paper, an optimized harmonic indicator function named enhanced probability statistics (EPS) was proposed to improve the effectiveness of harmonic component detection. Moreover, a new algorithm based on the harmonic decomposition was presented to remove the harmonic components before modal estimation. The presented method was validated by a simulation example of a plane frame and a practical modal test of the GARTUER plane model. Compared with the method based on experimental modal analysis without considering harmonic excitation, the new method offers better and more reliable parameter identification.