The human auditory system can effectively capture, screen and trace each frequency component in acoustical signal. However in vibration signal analysis, the judgment of harmonic component and its time varying is always an artificial job according to frequency spectrum or time-frequency spectrum. Based on the operating mechanism of auditory system, a method is proposed that can identify and extract harmonic frequency automatically. In this method, the basilar membrane band-pass filtering is carried out firstly, and then simulating the function of temporal frequency receptive field of auditory cortex and each time-frequency domain is calculated by 2D-DFT to extract amplitude and frequency information. A identification method of harmonic frequency is proposed, then the frequency of harmonic and random component can be distinguished and the harmonic frequency points can be extracted by means of synthesizing frequency information and result of lateral inhibition of amplitude information. At last, the harmonic frequency points are grouped and the result can reveal each main harmonic frequency and change with time. The results of simulation and experiments show that the proposed method can extract each harmonic frequency accurately in signal with the anti-noise-interference ability, and can obtain a certain effect for those harmonic components with weak amplitude.