One important characteristic of current and voltage signals is their dynamic behavior which is usually ignored by the traditional phasor measurement algorithm, and inaccurate phasor estimates will be attain by the traditional algorithm because of dynamic behavior such as power system oscillations. The derivatives of Taylor expansion are employed to express dynamic signals and their coefficients are regarded as the constraints for an optimization function, which will generate a sequence to modify the coefficient of the filter. Not only will the physical meaning of the derivative coefficients be described in detail within this paper, but also an efficient phasor estimate algorithm, that has good performance under dynamic conditions, will be proposed. Theoretical analysis, simulation and analysis of real data demonstrate that the performance of the algorithm proposed by this paper is superior to traditional phasor estimate algorithm based on a FIR filter in the presence of oscillations and frequency deviations.