Adaptive Short-Time Fractional Fourier Transform Used in Time-Frequency Analysis

In order to improve the time-frequency resolutions of short-time fractional Fourier transform, adaptive short-time fractional Fourier transform (ASTFRFT) method is used in this paper. The optimal order of ASTFRFT is given by maximizing kurtosis of signals in fractional domain, where the window width of ASTFRFT is searched by the maximal Shannon entropy of timefrequency distribution. Short-time fractional Fourier transform has lots of transform orders and its window width has several options. ASTFRFT selects its fractional transform orders wtih kurtosis, and its width window with Shannon entropy of time-frequency distribution. As to nonstationary signals, the experimental results reveal that the ASTFRFT has better effect than the short-time fractional Fourier transform with arbitrary fractional orders and with arbitrary window width. For multi-components signals, ASTFRFT can easier and more efficiently select the optimal orders and their suitable window width for short-time fractional Fourier transform, which can provide good time-frequency resolutions.