Abstract: With the wide adoption of phasor measurement unit（PMU）in energy management systems,research on dynamic state estimator for synchronous machine based on PMU is attracting more and more attention.Should there exist bad data,the effect of filtering would be seriously affected.First,an algorithm for dynamic state estimation based on unscented Kalman filter is described.But as the PMU data is of poor quality,to solve problem,the time-varying residual threshold is found by deriving the residual equation.Then,the position of bad data measuring point is determined by an iterative detection method.The corresponding measurement of the bad data is ruled out and repeated for correction of the estimation.Simulation results show that the proposed algorithm can effectively restrain the influence of bad data on state estimation.