Variable-order fractional numerical differentiation for noisy signals by wavelet denoising

In this paper, a numerical method is proposed to estimate the variable-order fractional derivatives of an unknown signal in noisy environment. Firstly, the wavelet denoising process is adopted to reduce the noise effect for the signal. Secondly, polynomials are constructed to fit the denoised signal in a set of overlapped subintervals of a considered interval. Thirdly, the variable-order fractional derivatives of these fitting polynomials are used as the estimations of the unknown ones, where the values obtained near the boundaries of each subinterval are ignored in the overlapped parts. Finally, numerical examples are presented to demonstrate the efficiency and robustness of the proposed method.