Optimal Parametric Estimation of Biased Sinusoidal Signals Using DREM.

This brief presents a novel optimal parametric estimation method of improving the reconstructing accuracy for biased sinusoidal signals. First, a second-order generalized integrator (SOGI) is used to construct a linear regression equation, whose unknown coefficient vector is a combination with the bias and frequency of the input signal. Next, the dynamic regression extension and mixing technique is employed to estimate the unknown parameters by solving this equation. Then, the particle swarm optimization algorithm simultaneously optimizes the parameters of SOGI and adaptive control gains. Finally, a comparison of the accuracy of frequency estimation and reconstruction ability illustrates the superiority of the proposed strategy.