Abstract: The conventional frequency tracking algorithm for power systems generally relies on adjusting the response speed or accuracy of the chosen power system model for applications to enhance the stability of power system under abnormal states.To reduce the impacts of harmonic and noise on frequency measurements,an adaptive fundamental component extraction and frequency tracking algorithm was presented.By pre-estimating the state of power grid,this algorithm obtained the estimated frequency and amplitude of voltage,which were used to update infinite impulse response filter coefficients in real time,thus adaptive fundamental wave extraction was achieved.In addition,a method to rapidly stabilize the performance of the fundamental wave filter when filter’s coefficients were switched was proposed.On the base of this,a robust extended Kalman filter was introduced to achieve accurate frequency tracking.Simulation and experimental results show that the new algorithm increases the speed and accuracy of response,effectively suppresses the impacts of noise on the frequency tracking results,significantly lowers the arithmetic complexity over the conventional method and satisfies the requirements of real time applications for power system.