With the limitation of the error function used in adaptive notch filter (ANF)

the convergence speed of the adaptive frequency estimation method is slow and the setting of initial iteration frequency value is also restrained.Especially for the signal whose frequency is close to 0 or

the frequency estimation accuracy is not satisfied and the stability of ANF is poor

so a new frequency estimation method of ANF based on new error function is proposed.Firstly

the reasons for low accuracy

slow convergence speed and poor stability of ANF frequency estimation methods are analyzed

and a new error function is proposed to improve the convergence speed;secondly

according to the reasons of the bias generated by ANF

a bias removing technique is adopted to improve the accuracy of the frequency estimation and the unbiased frequency estimation results are obtained

meanwhile the discrete kalman filtering is used to improve stability of ANF

then the performance analysis of ANF frequency estimation methods at steady state is also given;thirdly

the frequency estimation results of ANF are acquired.The impacts of the parameters on the accuracy of ANF for frequency estimation are discussed.Consequently

the calculated experiment results are provided to confirm the effectiveness and correctness of the algorithm.