Average Plain Gradient Based Indirect Frequency Estimation Using Adaptive Notch Filter

Abstract

The existing Adaptive Notch Filter (ANF)-based frequency estimation methods have problems of slow convergence speed, unstable error and limited selection of iterative initial values. In this paper, a frequency estimation method based on the indirect average plain gradient algorithm is proposed. It uses a new error function applied to the second-order adaptive Infinite Impulse Response (IIR) notch filter. The proposed error function is the mean value of the weighted average of the squares of the two signals outputted by Finite Impulse Response (FIR) and IIR sections of the ANF. It has better gradient characteristics, and arbitrary initial value can be selected for the following iterative calculation. The theory and simulation show that the proposed algorithm, compared with other indirect gradient algorithm, improves the convergence speed and estimation accuracy with little addition of the computation, while it is superior in varying-frequency signal tracking performance.