The basic concepts and theories of Hilbert-Huang-Transformation (HHT) for signal analysis is introduced and its implementation procedure is programmed. For the collected three orthogonal directions wind speed signals at a measuring point at Sutong Bridge site, Empirical Mode Decomposition (EMD) is conducted to extract the Intrinsic Mode Functions (IMFs). Hilbert transformation is performed on IMFs and its marginal amplitude spectra and amplitude spectra are achieved. The orthogonal exponential matrix that reflects the orthogonalities between the IMFs is obtained, and eigenvalue decomposition is carried out. The results show that the measured wind speed appears non-stationary nature, even for the same measuring point, for the wind speed at different directions, the dominant frequencies are distinct.