Most phasor measurement algorithms consider signal phasors as a steady-state model, so they are very sensitive to voltage amplitude and angle fluctuations that often occur in modern power grids. Taylor Weighted Least Squares (TWLS) algorithm relies on a dynamic phasor model at nominal frequency. The algorithm not only provide the phasor values but also give its derivatives to improve the dynamic monitoring of the power grid. On this basis, an improved TWLS method based on fundamental frequency is proposed. First, the fundamental frequency is obtained by nonlinear least squares method. Then, the derivation process of the TWLS algorithm based on the calculated fundamental frequency is introduced, and the window function, the length of the data window and Taylor polynomial order involved in the algorithm are analyzed and selected. Finally, different signal models and actual data are used to test the performance of the algorithm. The simulation results show that the accuracy of the improved TWLS algorithm meets the requirements of synchronized phasor measurement. This work is supported by National Natural Science Foundation for Young Scholars (No. 51505317) and Natural Science Foundation of Shanxi Province (No. 201601D102039).